In today’s world, characterised by digitally interconnected public spheres, disinformation poses a significant risk to the democratic common good and to trust in the state and democratic institutions. However, due to their complexity, AI models used to detect digital disinformation are ‘black boxes’ for users: the AI’s output is not transparent, and AI predictions cannot be directly validated. In particular, the patterns used by an AI model to detect digital disinformation remain hidden. This makes it difficult to harness the AI’s implicit knowledge for educational purposes.
The EKILED research project aims to investigate interactive XAI (Explainable Artificial Intelligence) approaches that reveal and present the implicit knowledge of AI models for detecting disinformation in a way that is educationally valuable for users. Specifically, the aim is to develop the EKILED demonstrator, which will serve as a reflection partner for pupils in recognising digital disinformation through interactive XAI approaches. The project is based on the protective strategy of providing education about digital disinformation whilst pupils are still at school and strengthening the relevant media and digital skills, in order to embed social resilience and trust in democracy at an early stage.
Consequently, the EKILED demonstrator can be understood as an interactive learning system that enables pupils to recognise and understand patterns of digital disinformation in a playful way. The demonstrator is to be made available nationwide so that teachers can be empowered, with minimal barriers to entry, to teach pupils how to deal with digital disinformation in an informed manner. In the long term, EKILED thus contributes to societal resilience and the strengthening of democracy, and serves as a guide for the responsible use of trustworthy and explainable AI in educational contexts.
Research partners: University of Bamberg, Karlsruhe Institute of Technology (KIT)
Development partner: DASU – Centre for Digitalisation, Analytics & Data Science, Ulm
Application partner: Eduversum GmbH
Project period: March 2026 – March 2029
The project is funded by the Federal Ministry of Research, Technology and Space under the funding scheme ‘Trust in Democracy and the State: Detecting and Countering Digital Disinformation’ and is coordinated by the University of Ulm.